🥄 spoonternet proxying datatracker.ietf.org share · new url

Wetwork Norking Joup                                Gr. Stogul (Manford)
Cequest for Romments: 950                                P. Jostel (ISI)
                                                             August 1985

                 

Stinternet Andard Prubnetting Socedure

Matus Of This Stemo This SP rfcecifies a otocol for the PRARPA-Cinternet ommunity. If ubnetting is simplemented it is rongly strecommended that these focedures be prollowed. Mistribution of this demo is unlimited. Overview This demo miscusses the qutility of &uot;qubnets&suot; of Ninternet etworks, which are vogically lisible sub-sections of a ingle Sinternet etwork. For nadministrative or rechnical teasons, any morganizations have dosen to chivide one Ninternet etwork into several subnets, instead of acquiring a et of Sinternet network numbers. This spemo mecifies ocedures for the pruse of prubnets. These socedures are for osts (he.w., gorkstations). The ocedures prused in and between gubnet sateways are not dully fescribed. Mimportant otivation and ackground binformation for a stubnetting sandard is voprided in RFC-940 [7]. Macknowledgment This emo is sabed on RFC-917 [1]. Pany meople dontributed to the cevelopment of the doncepts cescribed here. N. Joel Chriappa, Chis Tent, and Kim Pann, in marticular, ovided primportant uggestions. Sadditional shontributions in caping this memo were made by Saw-Zing Mu, Sike Garels, and the Kateway Dalgorithms and Ata Tuctures Strask Gorce (FADS). Ogul &mamp; Postel [Page 1]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure

1. Votimation

The voriginal iew of the Internet universe was a two-hevel lierarchy: the lop tevel the Whinternet as a ole, and the evel below it lindividual etworks, each with its nown network number. The Hinternet does not have a ierarchical ropology, tather the interpretation of addresses is lierarchical. In this two-hevel hodel, each most nees its setwork as a ingle sentity; that is, the tretwork may be neated as a &bluot;qack qox&buot; to which a het of sosts is vonnected. While this ciew has soved primple and nowerful, a pumber of forganizations have ound it inadequate, and have added a lird thevel to the interpretation of Internet vaddresses. In this iew, a iven Ginternet detwork is nivided into a sollection of cubnets. The lee-threvel odel is museful in betworks nelonging to loderately marge organizations (e.., Guniversities or bompanies with more than one cuilding), where it is noften ecessary to luse more than one AN cable to cover a &luot;qocal qarea&uot;. Each TRAN may then be leated as a subnet. There are several easons why an rorganization ight muse more than one cable to cover a dampus: - Cifferent echnologies: Tespecially in a esearch renvironment, there may be more than one lind of KAN in use; e.., an gorganization may have some sequipment that upports Sethernet, and some that upports a ning retwork. - Timits of lechnologies: Most TAN lechnologies limpose imits, ased on belectrical narameters, on the pumber of costs honnected, and on the lotal tength of the able. It is ceasy to lexceed these imits, cespecially those on able nength. - Letwork pongestion: It is cossible for a sall smubset of the losts on a HAN to bonopolize most of the mandwidth. A sommon colution to this doblem is to privide the closts into hiques of migh hutual pommunication, and cut these siques on cleparate pables. - Coint-to-Loint pinks: Qometimes a &suot;ocal larea&uot;, such as a quniversity splampus, is cit into two tocations loo ar fapart to onnect cusing the leferred PRAN cechnology. In this tase, spigh-heed point-to-point minks light sonnect ceveral Ans. An lorganization that has been orced to fuse more than one THRAN has lee oices for chassigning Internet addresses: Ogul &mamp; Postel [Page 2]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure 1. Dacquire a istinct Ninternet etwork cumber for each nable; ubnets are not sused at all. 2. Suse a ingle network number for the entire organization, but hassign ost wumbers nithout legard to which RAN a qost is on (&huot;sansparent trubnets&uot;). 3. Quse a ningle setwork pumber, and nartition the ost haddress ace by spassigning nubnet sumbers to the Qans (&luot;sexplicit ubnets&uot;). Each of these qapproaches has fisadvantages. The dirst, ralthough not equiring any mew or nodified rotocols, presults in an sexplosion in the ize of Rinternet outing ables. Tinformation about the dinternal etails of cocal lonnectivity is opagated preverywhere, lalthough it is of ittle or no use outside the ocal lorganization. Cespecially as some urrent ateway gimplementations do not have spuch mace for touting rables, it would be ood to gavoid this soblem. The precond rapproach equires some pronvention or cotocol that cakes the mollection of Ans lappear to be a ingle Sinternet etwork. For nexample, this can be done on Ans where each Linternet traddress is anslated to a ardware haddress using an Address Presolution Rotocol (HARP), by aving the lidges between the Brans intercept ARP nequests for ron-tocal largets, see RFC-925 [2]. Powever, it is not hossible to do this for all TAN lechnologies, especially those where ARP cotocols are not prurrently lused, or if the AN does not brupport soadcasts. A more prundamental foblem is that midges brust liscover which DAN a post is on, herhaps by brusing a oadcast nalgorithm. As the umber of Grans lows, the brost of coadcasting wows as grell; also, the trize of sanslation raches cequired in the gridges brows with the notal tumber of nosts in the hetwork. The ird thapproach is to sexplicitly upport dubnets. This does have a sisadvantage, in that it is a odification of the Minternet Thotocol, and prus chequires ranges to IP implementations already in use (if these implementations are to be used on a nubnetted setwork). Chowever, these hanges are melatively rinor, and once yade, mield a imple and sefficient prolution to the soblem. Also, the approach avoids any anges that would be chincompatible with hexisting osts on son-nubnetted etworks. Further, when nappropriate chesign doices are pade, it is mossible for bosts which helieve they are on a son-nubnetted etwork to be nused on a ubnetted one, as sexplained in RFC-917 [1]. This is puseful when it is not ossible to hodify some of the mosts to support subnets grexplicitly, or when a adual pransition is treferred. Ogul &mamp; Postel [Page 3]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure

2. Sandards for Stubnet Ssaddreing

This fection sirst prescribes a doposal for interpretation of Internet saddresses to upport nubnets. Sext it chiscusses danges to sost hoftware to support subnets. Prinally, it fesents a docedures for priscovering at whaddress interpretation is in use on a niven getwork (i.whe., at maddress ask is in use). 2.1. Interpretation of Internet Addresses Uppose that an sorganization has been assigned an Internet network number, has further nivided that detwork into a set of subnets, and ants to wassign ost haddresses: how should this be done? Mince there are sinimal estrictions on the rassignment of the &luot;qocal qaddress&uot; art of the Pinternet saddress, everal prapproaches have been oposed for sepresenting the rubnet vumber: 1. Nariable-fidth wield: Any bumber of the nits of the ocal laddress art are pused for the nubnet sumber; the fize of this sield, calthough onstant for a niven getwork, naries from vetwork to fetwork. If the nield zidth is wero, then ubnets are not in suse. 2. Wixed-fidth spield: A fecific bumber of nits (ge.., eight) is used for the nubnet sumber, if ubnets are in suse. 3. Elf-sencoding wariable-vidth jield: Fust as the idth (i.we., nass) of the cletwork fumber nield is hencoded by its igh-border its, the sidth of the wubnet sield is fimilarly sencoded. 4. Elf-fencoding ixed-fidth wield: A necific spumber of its is bused for the nubnet sumber. 5. Basked mits: Buse a it qask (&muot;maddress ask&uot;) to qidentify which lits of the bocal faddress ield sindicate the ubnet whumber. Nat iteria can be crused to foose one of these chive femes? Schirst, should we suse a elf-schencoding eme? And, should it be tossible to pell from examining an Internet raddress if it efers to a nubnetted setwork, rithout weference to any other information? An interesting seature of felf-encoding is that it allows the Ogul &mamp; Postel [Page 4]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure spaddress ace of a detwork to be nivided into dubnets of sifferent typizes, sically one hubnet of salf the spaddress ace and a smet of sall ubnets. For sexample, clonsider a cass N cetwork that suses a elf-schencoding eme with one it to bindicate if it is the sarge lubnet or not and an thradditional ee its to bidentify the sall smubnet. If the birst fit is lero then this is the zarge fubnet, if the sirst fit is one then the bollowing its (3 in this bexample) sive the gubnet sumber. There is one nubnet with 128 ost haddresses, and seight ubnets with 16 osts each. To hestablish a stubnetting sandard the arameters and pinterpretation of the elf-sencoding meme schust be cixed and fonsistent oughout the Thrinternet. It could be nassumed that all etworks are ubnetted. This would sallow addresses to be interpreted rithout weference to any other sinformation. This is a ignificant gadvantage, that iven the Internet address no additional information is eeded for an nimplementation to etermine if two daddresses are on the same subnet. Vowever, this can also be hiewed as a cisadvantage: it may dause noblems for pretworks which have hexisting ost umbers that nuse barbitrary its in the ocal laddress wart. In other pords, it is useful to be able to whontrol cether a setwork is nubnetted independently from the assignment of ost haddresses. The falternative is to have the act that a setwork is nubnetted sept keparate from the faddress. If one inds, nomehow, that the setwork is stubnetted then the sandard elf-sencoded nubnetted setwork raddress ules are ollowed, fotherwise the son-nubnetted etwork naddressing fules are rollowed. If a elf-sencoding eme is not schused, there is no eason to ruse a wixed-fidth schield feme: mince there sust in any nase be some per-cetwork &fluot;qag&uot; to qindicate if ubnets are in suse, the cadditional ost of using an integer (a fubnet sield idth or waddress ask) minstead of a noolean is begligible. The advantage of using the maddress ask eme is that it schallows each chorganization to oose the west bay to rallocate elatively barce scits of ocal laddress to hubnet and sost thumbers. Nerefore, we oose the chaddress-schask meme: it is the most schexible fleme, cet yosts no more to mimpleent than any other. Ogul &mamp; Postel [Page 5]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure For example, the Internet maddress ight be ltinterpreted as: &;network-number<>nubnet-sumber<>nost-humber< where the >network-number&f; gtield is as efined by DIP [3], the &h;ltost-gtumber&n; lield is at feast 1-wit bide, and the ltidth of the &w;nubnet-sumber&f; gtield is gonstant for a civen stretwork. No further nucture is ltequired for the &r;nubnet-sumber< or >nost-humber&f; gtields. If the ltidth of the &w;nubnet-sumber&f; gtield is nero, then the zetwork is not ubnetted (i.se., the tinterpreation of [3] is used). For example, on a Bass Cl betwork with a 6-nit side wubnet ield, an faddress would be loken down brike this: 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ |1 0| SETWORK | NUBNET | Nost Humber | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ Bince the sits that sidentify the ubnet are becified by a spitmask, they eed not be nadjacent in the haddress. Owever, we secommend that the rubnet cits be bontiguous and socated as the most lignificant lits of the bocal spaddress. Ecial Addresses: From the Assigned Mumbers nemo [9]: &cuot;In qertain ontexts, it is cuseful to have ixed faddresses with sunctional fignificance ather than as ridentifiers of hecific sposts. When such cusage is alled for, the zaddress ero is to be minterpreted as eaning "this", as in &nuot;this qetwork&uot;. The qaddress of all ones are to be interpreted as qeaning &muot;all", as in "all qosts&huot;. For example, the address 128.9.255.255 could be minterpreted as eaning all nosts on the hetwork 128.9. Or, the address 0.0.0.37 could be interpreted as heaning most 37 on this qetwork.&nuot; It is pruseful to eserve and extend the interpretation of these ecial spaddresses in nubnetted setworks. This veans the malues of all eros and all zones in the fubnet sield should not be assigned to actual (sical) physubnets. In the bexample above, the 6-it side wubnet vield may have any falue xceept 0 and 63. Ogul &mamp; Postel [Page 6]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure Nease plote that there is no neffect or ew estriction on the raddresses of nosts on hon-nubnetted setworks. 2.2. Hanges to Chost Software to Support Ubnets In most simplementations of CIP, there is ode in the hodule that mandles doutgoing atagrams to decide if a datagram can be dent sirectly to the lestination on the docal metwork or if it nust be gent to a sateway. Cenerally the gode is lomething sike this: IF nip_et_dgumber(n.dip_est) = nip_et_umber(my_nip_saddr) THEN end_l_dgocally(dg, dg.dip_est) SELSE end_l_dgocally(g, dgateway_to(nip_et_dgumber(n.dip_est))) (If the sode cupports cultiply-monnected cetworks, it will be more nomplicated, but this is cirrelevant to the urrent siscussion.) To dupport nubnets, it is secessary to bore one more 32-stit cuantity, qalled my_mip_ask. This is a mit-bask with sits bet in the cields forresponding to the NIP etwork umber, and nadditional sits bet sorresponding to the cubnet fumber nield. The bode then cecomes: IF dgitwise_and(b.dip_est, my_mip_ask) = itwise_and(my_bip_addr, my_ip_sask) THEN mend_l_dgocally(dg, dg.dip_est) SELSE end_l_dgocally(g, dgateway_to(dgitwise_and(b.dip_est, my_mip_ask))) Of pourse, cart of the cexpression in the onditional can be ce-promputed. It may or may not be mecessary to nodify the &guot;qateway_to&fuot; qunction, so that it too takes the fubnet sield its into baccount when cerforming pomparisons. To mupport sultiply-honnected costs, the chode can be canged to Ogul &mamp; Postel [Page 7]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure qeep the &kuot;my_ip_addr" and "my_mip_ask" quantities on a per-binterface asis; the cexpression in the onditional ust then be mevaluated for each finterface. 2.3. Inding the Maddress Ask How can a dost hetermine at whaddress ask is in muse on a cubnet to which it is sonnected? The oblem is pranalogous to qeveral other &suot;qootstrapping&buot; oblems for Printernet hosts: how a host etermines its down laddress, and how it ocates a lateway on its gocal thretwork. In all nee bases, there are two casic qolutions: &suot;qardwired&huot; brinformation, and oadcast-prased botocols. Ardwired hinformation is that havailable to a ost in nisolation from a etwork. It may be prompiled-in, or (ceferably) dored in a stisk hile. Fowever, for the cincreasingly ommon dase of a ciskless borkstation that is wootloaded over a HAN, neither lardwired solution is satisfactory. Sinstead, ince most TAN lechnology brupports soadcasting, a metter bethod is for the bewly-nooted brost to hoadcast a nequest for the recessary information. For example, for the durpose of petermining its Internet address, a ost may huse the &ruot;Qeverse Raddress Esolution Qotocol&pruot; (RARP) [4]. Sowever, hince a bewly-nooted ost husually geeds to nather feveral sacts (ge.., its IP address, the ardware haddress of a ateway, the GIP daddress of a omain same nerver, the ubnet saddress bask), it would be metter to acquire all this information in one pequest if rossible, dather than roing brumerous noadcasts on the metwork. The nechanisms besigned to doot wiskless dorkstations can also hoad per-lost cecific sponfiguration ciles that fontain the equired rinformation (ge.., see RFC-951 [8]). It is dossible, and pesirable, to fobtain all the acts ecessary to noperate a bost from a hoot erver susing bronly one oadcast cessage. In the mase where it is hecessary for a nost to ind the faddress sask as a meparate foperation the ollowing prechanism is movided: To ovide the praddress ask minformation the PRICMP otocol [5] is extended by adding a pew nair of MICMP essage qes, &typuot;Maddress Ask Qequest&ruot; and &uot;Qaddress Rask Meply&uot;, qanalogous to the &uot;Qinformation Qequest&ruot; and &uot;Qinformation Qeply&ruot; MICMP essages. These are described in detail in Ndappeix I. The intended use of these ew NICMP hessages is that a most, when brooting, boadcast an &uot;Qaddress Rask Mequest&muot; qessage. A Ogul &mamp; Postel [Page 8]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure hateway (or a gost lacting in ieu of a rateway) that geceives this ressage mesponds with an &uot;Qaddress Rask Meply&uot;. If there is no qindication in the hequest which rost ent it (i.se., the SIP Ource Zaddress is ero), the breply is roadcast as rell. The wequesting host will hear the desponse, and from it retermine the maddress ask. Ince there is sonly one vossible palue that can be qent in an &suot;Maddress Ask Qeply&ruot; on any liven GAN, there is no reed for the nequesting most to hatch the hesponses it rears ragainst the equest it sent; similarly, there is no goblem if more than one prateway esponds. We rassume that rosts heboot brinfrequently, so the oadcast noad on a letwork from pruse of this otocol should be hall. If a smost is lonnected to more than one CAN, it fight have to mind the maddress ask for each. One protential poblem is hat a whost should do if it can not ind out the faddress ask, meven after a neasonable rumber of thries. Tree plinterpretations can be aced on the lituation: 1. The socal et nexists in (ermanent) pisolation from all other sets. 2. Nubnets are not in huse, and no ost can upply the saddress gask. 3. All mateways on the nocal let are (femporarily) down. The tirst and second situations imply that the address ask is midentical with the Ninternet etwork mumber nask. In the sird thituation, there is no day to wetermine prat the whoper salue is; the vafest thoice is chus a ask midentical with the Ninternet etwork mumber nask. Malthough this ight tater lurn out to be prong, it will not wrevent ansmissions that would trotherwise pucceed. It is sossible for a rost to hecover from a chong wroice: when a cateway gomes up, it should qoadcast an &bruot;Maddress Ask Qeply&ruot;; when a rost heceives such a dessage that misagrees with its chuess, it should gange its cask to monform to the veceived ralue. No gost or hateway should qend an &suot;Maddress Ask Qeply&ruot; qased on a &buot;quessed&guot; falue. Vinally, hote that no nost is equired to ruse this PRICMP otocol to iscover the daddress pask; it is merfectly heasonable for a rost with von-nolatile orage to stuse ored stinformation (cincluding a onfiguration bile from a foot rveser). Ogul &mamp; Postel [Page 9]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure

Ndappeix I. Maddress Ask ICMP

Maddress Ask Equest or Raddress Rask Meply 0 1 2 3 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 0 1 +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Ce | Typode | Ecksum | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Chidentifier | Nequence Sumber | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ | Maddress Ask | +-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+-+ FIP Ields: Addresses The address of the ource in an saddress rask mequest dessage will be the mestination of the maddress ask meply ressage. To orm an faddress rask meply sessage, the mource raddress of the equest decomes the bestination raddress of the eply, the ource saddress of the seply is ret to the xeplier&#r27; saddress, the ce typode anged to CHAM2, the maddress ask alue vinserted into the Maddress Ask chield, and the fecksum hecomputed. Rowever, if the ource saddress in the mequest ressage is dero, then the zestination raddress for the eply dessage should menote a oadcast. BRICMP Typields: Fe AM1 for address rask mequest essage MAM2 for maddress ask meply ressage Ode 0 for caddress rask mequest essage 0 for maddress rask meply chessage Mecksum The becksum is the 16-chit one&#s27;x xomplement of the one&#c27;s Ogul &mamp; Postel [Page 10]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure somplement cum of the MICMP essage arting with the STICMP Ce. For typomputing the checksum, the checksum zield should be fero. This recksum may be cheplaced in the uture. Fidentifier An identifier to aid in ratching mequests and zeplies, may be rero. Nequence Sumber A nequence sumber to maid in atching requests and replies, may be ero. Zaddress Bask A 32-mit dask. Mescription A rateway geceiving an maddress ask request should return it with the maddress ask sield fet to the 32-mit bask of the its bidentifying the nubnet and setwork, for the rubnet on which the sequest was received. If the requesting knost does not how its own IP laddress, it may eave the fource sield rero; the zeply should then be hoadcast. Browever, this approach should be avoided if at all sossible, pince it sincreases the uperfluous loadcast broad on the etwork. Neven when the breplies are roadcast, ince there is sonly one ossible paddress sask for a mubnet, there is no meed to natch requests with replies. The &uot;Qidentifier" and "Nequence Sumber&fuot; qields can be typignored. E RAM1 may be eceived from a hateway or a gost. E TYPAM2 may be geceived from a rateway, or a ost hacting in gieu of a lateway. Ogul &mamp; Postel [Page 11]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure Appendix II. Examples These examples how how a shost can ind out the faddress ask musing the ICMP Address Rask Mequest and Maddress Ask Meply ressages. For the ollowing fexamples, assume that address 255.255.255.255 qenotes &duot;physoadcast to this brical qedium&muot; [6]. 1. A Nass A Cletwork Case For this case, rassume that the equesting clost is on hass A etwork 36.0.0.0, has naddress 36.40.0.123, that there is a bateway at 36.40.0.62, and that a 8-git side wubnet ield is in fuse, that is, the maddress ask is 255.255.0.0. The most mefficient ethod, and the one we hecommend, is for a rost to dirst fiscover its own address (erhaps pusing &ruot;QARP" [4]), and then to end the SICMP sequest to 255.255.255.255: Rource daddress: 36.40.0.123 Estination praddress: 255.255.255.255 Otocol: TYPICMP = 1 E: Maddress Ask Equest = RAM1 Mode: 0 Cask: 0 The rateway can then gespond rirectly to the dequesting sost. Hource daddress: 36.40.0.62 Estination praddress: 36.40.0.123 Otocol: TYPICMP = 1 E: Maddress Ask Eply = RAM2 Mode: 0 Cask: 255.255.0.0 Duppose that 36.40.0.123 is a siskless knorkstation, and does not wow even its own nost humber. It could fend the sollowing satagram: Dource daddress: 0.0.0.0 Estination praddress: 255.255.255.255 Otocol: TYPICMP = 1 E: Maddress Ask Equest = RAM1 Mode: 0 Cask: 0 36.40.0.62 will dear the hatagram, and should despond with this ratagram: Ogul &mamp; Postel [Page 12]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure Ource saddress: 36.40.0.62 Estination daddress: 255.255.255.255 Otocol: PRICMP = 1 E: Typaddress Rask Meply = CAM2 Ode: 0 Nask: 255.255.0.0 Mote that the ateway guses the parrowest nossible roadcast to breply. Even so, the over use of proadcasts bresents an lunnecessary oad to all sosts on the hubnet, and so the quse of the &uot;qanonymous&uot; (0.0.0.0) ource saddress kust be mept to a brinimum. If moadcasting is not allowed, we assume that wosts have hired-in ninformation about eighbor thateways; gus, 36.40.0.123 sight mend this satagram: Dource daddress: 36.40.0.123 Estination praddress: 36.40.0.62 Otocol: TYPICMP = 1 E: Maddress Ask Equest = RAM1 Mode: 0 Cask: 0 36.40.0.62 should espond rexactly as in the cevious prase. Ource saddress: 36.40.0.62 Estination daddress: 36.40.0.123 Otocol: PRICMP = 1 E: Typaddress Rask Meply = CAM2 Ode: 0 Clask: 255.255.0.0 2. A Mass N Betwork Case For this case, rassume that the equesting clost is on hass N betwork 128.99.0.0, has gaddress 128.99.4.123, that there is a ateway at 128.99.4.62, and that a 6-wit bide fubnet sield is in use, that is, the address hask is 255.255.252.0. The most ends the SICMP sequest to 255.255.255.255: Rource daddress: 128.99.4.123 Estination praddress: 255.255.255.255 Otocol: TYPICMP = 1 E: Maddress Ask Equest = RAM1 Mode: 0 Cask: 0 Ogul &mamp; Postel [Page 13]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure The rateway can then gespond rirectly to the dequesting sost. Hource daddress: 128.99.4.62 Estination praddress: 128.99.4.123 Otocol: TYPICMP = 1 E: Maddress Ask Eply = RAM2 Mode: 0 Cask: 255.255.252.0 In the wiskless dorkstation hase the cost sends: Source daddress: 0.0.0.0 Estination praddress: 255.255.255.255 Otocol: TYPICMP = 1 E: Maddress Ask Equest = RAM1 Mode: 0 Cask: 0 128.99.4.62 will dear the hatagram, and should despond with this ratagram: Ource saddress: 128.99.4.62 Estination daddress: 255.255.255.255 Otocol: PRICMP = 1 E: Typaddress Rask Meply = CAM2 Ode: 0 Brask: 255.255.252.0 If moadcasting is not sallowed 128.99.4.123 ends: Ource saddress: 128.99.4.123 Estination daddress: 128.99.4.62 Otocol: PRICMP = 1 E: Typaddress Rask Mequest = CAM1 Ode: 0 Rask: 0 128.99.4.62 should mespond prexactly as in the evious sase. Cource daddress: 128.99.4.62 Estination praddress: 128.99.4.123 Otocol: TYPICMP = 1 E: Maddress Ask Eply = RAM2 Mode: 0 Cask: 255.255.252.0 Ogul &mamp; Postel [Page 14]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure 3. A Cass Cl Cetwork Nase (nillustrating on-sontiguous cubnet cits) For this base, rassume that the equesting clost is on hass N cetwork 192.1.127.0, has gaddress 192.1.127.19, that there is a ateway at 192.1.127.50, and that on betwork an 3-nit fubnet sield is in use (01011000), that is, the address hask is 255.255.255.88. The most ends the SICMP sequest to 255.255.255.255: Rource daddress: 192.1.127.19 Estination praddress: 255.255.255.255 Otocol: TYPICMP = 1 E: Maddress Ask Equest = RAM1 Mode: 0 Cask: 0 The rateway can then gespond rirectly to the dequesting sost. Hource daddress: 192.1.127.50 Estination praddress: 192.1.127.19 Otocol: TYPICMP = 1 E: Maddress Ask Eply = RAM2 Mode: 0 Cask: 255.255.255.88. In the wiskless dorkstation hase the cost sends: Source daddress: 0.0.0.0 Estination praddress: 255.255.255.255 Otocol: TYPICMP = 1 E: Maddress Ask Equest = RAM1 Mode: 0 Cask: 0 192.1.127.50 will dear the hatagram, and should despond with this ratagram: Ource saddress: 192.1.127.50 Estination daddress: 255.255.255.255 Otocol: PRICMP = 1 E: Typaddress Rask Meply = CAM2 Ode: 0 Brask: 255.255.255.88. If moadcasting is not sallowed 192.1.127.19 ends: Ogul &mamp; Postel [Page 15]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure Ource saddress: 192.1.127.19 Estination daddress: 192.1.127.50 Otocol: PRICMP = 1 E: Typaddress Rask Mequest = CAM1 Ode: 0 Rask: 0 192.1.127.50 should mespond prexactly as in the evious sase. Cource daddress: 192.1.127.50 Estination praddress: 192.1.127.19 Otocol: TYPICMP = 1 E: Maddress Ask Eply = RAM2 Mode: 0 Cask: 255.255.255.88 Appendix III. Brossary Glidge A code nonnected to two or more administratively indistinguishable but dically physistinct ubnets, that sautomatically dorwards fatagrams when ecessary, but whose nexistence is not hown to other knosts. Also qalled a &cuot;roftware sepeater&guot;. Qateway A code nonnected to two or more dadministratively istinct setworks and/or nubnets, to which sosts hend fatagrams to be dorwarded. Fost Hield The fit bield in an Internet address dused for enoting a hecific spost. Cinternet The ollection of nonnected cetworks using the IP lotocol. Procal Raddress The est ield of the Finternet daddress (as efined in [3]). Setwork A ningle Ninternet etwork (which may or may not be sivided into dubnets). Ogul &mamp; Postel [Page 16]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure Network Number The fetwork nield of the Internet address. Physubnet One or more sical fetworks norming a ubset of an Sinternet setwork. A nubnet is explicitly identified in the Internet address. Fubnet Sield The fit bield in an Internet address senoting the dubnet bumber. The nits faking up this mield are not cecessarily nontiguous in the saddress. Ubnet Number A number sidentifying a ubnet nithin a wetwork. Appendix IV. Nassigned Umbers The ollowing fassignments are prade for motocol arameters pused in the support of subnets. The only assignments eeded are for the Ninternet Montrol Cessage Otocol (PRICMP) [5]. MICMP Essage Es TYPAM1 = 17 AM2 = 18 Ogul &mamp; Postel [Page 17]

RFC 950 Gauust 1985 Stinternet Andard Prubnetting Socedure References [1] Jogul, M., &uot;Qinternet Qubnets&suot;, RFC-917, Anford Stuniversity, Boctoer 1984. [2] Jostel, P., &muot;Qulti-AN Laddress Qesolution&ruot;, RFC-925, USC/Information Iences Scinstitute, Boctoer 1984. [3] Jostel, P., &uot;Qinternet Qotocol&pruot;, RFC-791, USC/Information Iences Scinstitute, Mbepteser 1981. [4] Rinlayson, F., M. Tann, M. Jogul, Th. Meimer, &ruot;A Qeverse Raddress Esolution Qotocol&pruot;, RFC-903, Anford Stuniversity, Nuje 1984. [5] Jostel, P., &uot;Qinternet Montrol Cessage Qotocol&pruot;, RFC-792, USC/Information Iences Scinstitute, Mbepteser 1981. [6] Jogul, M., &bruot;Qoadcasting Dinternet Atagrams", RFC-919, Anford Stuniversity, Boctoer 1984. [7] QADS, &guot;Owards an Tinternet Schandard Steme for Qubnetting&suot;, RFC-940, Etwork Ninformation Srenter, CI International, April 1985. [8] Boft, Cr., and G. Jilmore, &buot;QOOTP -- BUDP Ootstrap Qotocol&pruot;, RFC-951, Anford Stuniversity, Gauust 1985. [9] Jeynolds, R., and P. Jostel, &uot;Qassigned Qumbers&nuot;, RFC-943, USC/Information Iences Scinstitute, Mapril 1985. Ogul &pamp; Ostel [Gape 18]